CD150 Association with Either the SH2-Containing Inositol Phosphatase or the SH2-Containing Protein Tyrosine Phosphatase Is Regulated by the Adaptor Protein SH2D1A

Abs and reagents

Rabbit antisera against SHIP, SHP-2, and p38 mitogen-activated protein kinase; were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Glutathione-agarose was purchased from Sigma (St. Louis, MO, protein A- and protein G-Sepharose were from Pharmacia (Piscataway, NJ). For cross-linking of surface receptors, we used the following mAbs: IPO-3 anti-CD150 (IgG1); IPO-4 anti-CD95 (IgM) (22); and G28-5 anti-CD40 (IgG1) (23). F(ab′)₂ of goat anti-human IgM (Jackson ImmunoResearch, West Grove, PA) were used for IgM cross-linking on human B cell lines. To generate a polyclonal Ab recognizing SH2D1A (DSHP), one of us (K.E.N.) immunized rabbits with a 26-aa peptide conjugated to keyhole limpet hemocyanin (peptide sequence: EKKSSARSTQGTTGIREDPDVCLKAP). After three injections, serum was collected and the anti-SH2D1A titer was determined using an ELISA with free peptide bound in the solid phase (Research Genetics, Huntsville, AL).

Plasmid constructs

The GST-fusion protein construct of the cytoplasmic tail of CD150 (GST-CD150ct) was prepared as described (5). Forward and reverse primers with the appropriate restriction sites for in-frame cloning into the pGEX-2T plasmid were used to amplify the cDNA fragments using pfu polymerase (Stratagene, La Jolla, CA). Using PCR-based site-directed mutagenesis (24), we made constructs of GST-CD150ct fusion proteins with phenylalanine (F) replacements at tyrosines Y269, Y281, Y307, and Y327. Plasmids with the correct nucleotide sequences were transformed into the bacterial strain XLI-BlueMRF′ (Stratagene) for fusion protein production. Plasmids containing GST-CD150ct were also transformed into the Escherichia coli strain TKX1 (Stratagene) for production of tyrosine-phosphorylated fusion proteins. Tyrosine phosphorylation of these fusion proteins apparently was restricted only to the corresponding cytoplasmic tails, because GST was not tyrosine phosphorylated when expressed alone in the same bacteria. Expression and purification of GST fusion proteins were performed as described (25).

SH2D1A sequence analysis

DNA was extracted from EBV-immortalized cell lines according to standard protocols. The SH2D1A coding sequence was PCR amplified using Expand Taq polymerase (Boehringer Mannheim, Indianapolis, IN) and primers flanking each of the four SH2D1A exons as described (13).

Cell lines and stimulation

The pre-B cell lines REH and Namalwa; Burkitt’s lymphoma cell lines Ramos, BJAB, and Raji; the B lymphoma line B104; the B lymphoblastoid cell lines (B-LCL) CESS, MP-1, T5-1, 6.16, and RPMI-1788; and the Jurkat T cell line were maintained as described (26). B-LCL from XLP patients included: IARC 739 (interstitial deletion of SH2D1A), XLP-D (C→T mutation at the position 462), XLP-8005 (C→T mutation at the position 471), XLP-8002 (no mutations in SH2D1A) (13). T cell-depleted tonsillar cells were prepared as described (3, 25).

Cell staining

For determination of cell surface phenotype and cytoplasmic expression of SH2D1A, cells were surface stained with biotin-labeled anti-IgD, anti-CD150, or anti-CD95, followed by streptavidin-PerCP (Becton Dickinson, Mountain View, CA) and PE-labeled anti-CD38 (PharMingen, San Diego, CA), anti-CD3, or anti-CD20 (Becton Dickinson). Cells were fixed in 1% paraformaldehyde for 20 min and permeabilized with 0.2% Tween 20 for 15 min. Then rabbit anti-SH2D1A serum (affinity-purified using SH2D1A peptide) was added followed by goat F(ab′)₂ anti-rabbit IgG. Cells were washed twice with PBS containing 2.5% FCS and 0.2% Tween 20 and then analyzed on a FACScan flow cytometer (Becton Dickinson).

Biochemical methods

Cell lysis, immunoprecipitation, SDS-PAGE, in vitro kinase assays, and subcellular fractionations were performed as described (3, 27, 28). Western blotting was performed with an ECL kit (Amersham, Arlington Heights, IL). For evaluation of kinase activities, immunoprecipitates were washed with Nonidet P-40 (NP-40) lysis buffer or with NP-40 lysis buffer containing 0.5 M NaCl, twice with high salt buffer (0.5 M LiCl), and once with NP-40 lysis buffer and were subjected to in vitro kinase assays.

Modeling of the CD150 cytoplasmic tail

Preliminary CD150 homologue searches were performed with BLAST and PSI-BLAST over various protein databases including TrEMBL, Swiss-Prot, Kabad, and PDB. CD150 homologues were aligned with the ClustalX program. Program alignment output was slightly modified manually with JalView and GeneDoc alignment viewers to align TxYxxV/I-containing regions. We used HMMER package tools (29, 30, 31) to build hidden Markov model (HMM) profiles for the multiple alignment of CD150, 2B4, CD84, and Ly9 and to search TrEMBL, Kabat, and Swiss-Prot databases using this profile as a query. As the number of identified homologues grew, new sequences were added to the multiple alignments, and new HMM profiles were built. Identification of CD150ct structural homologues was done by both sequence (BLAST, Fasta, CPHmodels, UCSC HMM) and structure alignment H3P2 at UCLA-Department of Energy (32, 33). CD150 sequence fitting to the recognized fold was performed in a Swiss PDB viewer (34, 35) and submitted for modeling to the Swiss-Model server (http://www.expasy.ch/swissmod/SWISS-MODEL.html). The quality of the modeled structure was assessed by ERRAT (36) and Verify3D (37) programs. We used PovRay ray tracing software to prepare the presented picture.

SH2D1A is expressed in B cells where it associates with CD150

CD150 is expressed on the surface of B cells and is up-regulated after activation (3). Using immunohistology and a microarray analysis, CD150 was found in diffuse large B cell lymphoma (22, 38). Because it was shown that in T cells CD150 associates with SH2D1A, we tested whether SH2D1A is expressed in B cells and whether it associates with CD150 in B cells. We assess the expression of SH2D1A protein in B lineage cells in a panel of B cell lines representing different stages of maturation. All studied B-LCL, including cell lines from patients with XLP (IARC 739, XLP-D, XLP-8002, XLP-8005), expressed high levels of CD150 peak fluorescence intensity, >10.0). The BL lines Raji, Namalwa (EBV⁺), and BJAB (EBV⁻) expressed CD150 at a moderate level (peak fluorescence intentisy 7.0); all other B cell lines tested (REH, Ramos, B104, RPMI-8226) were CD150 negative. Western blot analysis of whole cell lysates revealed that SH2D1A was expressed in only two of the B lymphoblastoid cell lines studied, MP-1 and CESS (Fig. 1⇓A). Flow cytometry also showed intracellular expression of SH2D1A in MP-1, but not in the BJAB cell line (Fig. 1⇓B).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 1.

Expression of SH2D1A in B cells. A, The presence of SH2D1A in B-LCL lines MP-1 and CESS was revealed by Western blotting with an anti-SH2D1A serum. NP-40 lysates of 5 × 10⁵ cells/well were resolved in 15% SDS-PAGE. Positive control was the T cell line Jurkat. One of four experiments. B, Expression of SH2D1A in B cell lines BJAB and MP-1. Rabbit IgG served as a negative control for purified anti-SH2D1A rabbit Ab. C, Expression of SH2D1A in tonsillar B cells. Cells were stained for IgD, CD38, or CD150 surface expression and cytoplasmic SH2D1A using three-color flow cytometry. Rabbit IgG was a negative control for purified anti-SH2D1A Ab. Naive IgD⁺CD38⁻ B cells; mean, 85.4. Germinal center IgD⁻CD38⁺ B cells, SH2D1A⁺ cells, 32.0%; mean, 94.4; peak channel (PKchan), 112.0. CD38⁺CD150⁺ cells, SH2D1A⁺ cells, 46.14%; mean, 101.5; PKchan, 119.0. IgD⁻CD38⁻ cells, mean 70.3. One of four experiments.

The surface phenotype of MP-1 cells resembles early germinal center cells: IgM⁺IgD⁻CD38⁺CD39^lowCD95⁺CD150⁺. The CESS cell line has a surface phenotype similar to memory B cells; e.g., it expresses membrane IgG, but not IgM, and is CD39^high. In agreement with prior studies (13), immunohistochemical analysis revealed SH2D1A expression in frozen sections from tonsils, lymph nodes, and diffuse large B cell lymphoma. SH2D1A was present within the cytoplasm of scattered large cells localized to B cell areas (data not shown). To evaluate the SH2D1A expression in subsets of B cells in vivo, we performed three-color analysis of T cell-depleted tonsillar cells (2.8 ± 0.3% of CD3⁺ lymphocytes). These studies revealed SH2D1A in 11.5–18.0% of CD20⁺ tonsillar B cells. Low levels of SH2D1A expression were found in naive IgD⁺CD38⁻B cells (mean, 85.4). SH2D1A was detected in 32.0% of germinal center IgD⁻CD38⁺ (9% of total B cells; mean, 94.4; peak, 112.0). The highest level of SH2D1A protein was detected in 46.1% of CD38⁺CD150⁺ cells (mean, 101.5; peak, 119.0) (Fig. 1⇑C). At the same time IgD⁻CD38⁻ cells, which include memory B cells, were SH2D1A negative (mean, 70.3) (Fig. 1⇑C). These data agree with immunohistochemical studies of frozen tonsillar sections that did not reveal preferential localization of SH2D1A in any B cell zone, but did detect the highest level of SH2D1A expression within large cells with cleaved nuclei localized in germinal centers (data not shown).

Association of CD150 with SHIP, SH2D1A, or SHP-2 in B cells

Because our data implied that CD150 and SH2D1A are coexpressed in B cells, we tested whether these molecules associate in B cells. In both SH2D1A⁺ cell lines (MP-1 and CESS), SH2D1A coprecipitated with CD150. As expected, SH2D1A was not detected with CD150 in SH2D1A⁻ lines T5-1, IARC 739, XLP-8005, or BJAB (Fig. 2⇓A). In some B cell lines, CD150 coprecipitated with SHIP, and a tyrosine-phosphorylated fusion protein of the CD150ct can bind SHP-2 (5). Also in COS-7 and mouse T cells, CD150 binds SHP-2, and this association can be blocked by SH2D1A (10, 11). Immunoprecipitation experiments followed by Western blot analysis clearly showed that in the SH2D1A-expressing cell lines MP-1 and CESS, CD150 coprecipitated with SHIP, and not SHP-2 (Fig. 2⇓A). In contrast, in all SH2D1A-negative cell lines tested, CD150 associated only with SHP-2 (Fig. 2⇓A). Trace amounts of SHP-2 were detected together with CD150 in the CESS cell line that has a much lower level of SH2D1A expression than the MP-1 line. Apparently, this differential binding of SHIP with SH2D1A vs SHP-2 did not depend on CD150 tyrosine phosphorylation, since CD150 is constitutively phosphorylated on tyrosine in both the SH2D1A⁺ and SH2D1A⁻ cell lines studied (Fig. 2⇓B). To evaluate a possible role for differential phosphorylation of any of four tyrosine residues in CD150ct, we used tyrosine-phosphorylated GST-fusion proteins of CD150ct (5). In the absence of SH2D1A in cell lysates (cell line BJAB), this fusion protein precipitated SHP-2, but in the presence of SH2D1A (cell lysates from the MP-1 cell line), GST-CD150ctPY bound not only SHP-2 but also SHIP (Fig. 2⇓C). Thus, the presence of SH2D1A-facilitated binding of SHIP to CD150. GST-CD150ctPY coprecipitation with both SHIP and SHP-2, apparently depends on the level of SH2D1A association with CD150. Preferential SHIP binding to the native CD150 molecules in SH2D1A⁺ cell lines may reflect associations in the context of intracellular localization.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 2.

A, Coprecipitation of CD150 with SHP-2 vs SHIP and SH2D1A. CD150 was immunoprecipitated with mAb IPO-3 directly coupled to Sepharose, and MOPC 21 myeloma protein directly coupled to Sepharose was used as a negative control; 50 × 10⁶ cells/immunoprecipitation. Western blot analysis with Abs against SHIP, SHP-2, and SH2D1A was performed using ECL. Whole cell lysates served as a positive control. All cell lines expressed CD150 at comparable levels. One of four experiments. B, Tyrosine phosphorylation of CD150 revealed in Western blot with mAb 4G10. CD150 was immunoprecipitated with mAb IPO-3 as described in A. C, Presence of SH2D1A-facilitated SHIP association with CD150. Nonphosphorylated and tyrosine-phosphorylated GST-fusion proteins of CD150ct were used for precipitations from MP-1 cells followed by Western blot with anti-SHIP, anti-SHP-2, and anti-SH2D1A serum. Tyrosine phosphorylation of fusion proteins was controlled with mAb 4G10 (Anti-PY).

Short term ligation of IgM on SH2D1A⁺ MP-1 cells did not change the level of SH2D1A and SHP-2 coprecipitated with CD150 but did reduce the level of SHIP associated with CD150 (Fig. 3⇓A). At the same time, the level of precipitable CD150 remained constant (Fig. 3⇓A). Under the same conditions, SHIP and SH2D1A levels in the cytosolic and particulate cell fractions remained constant (Fig. 3⇓B). We did not detect BCR-induced relocalization of either SHIP or SH2D1A to detergent-insoluble glycolipid-enriched domains (Fig. 3⇓B). In contrast, IgM cross-linking had no effect on the level of either SHIP or SHP-2 association with CD150 in the SH2D1A⁻ cell line BJAB (Fig. 3⇓A, lower panels) or XLP B-LCLs (data not shown). This suggests that BCR ligation can alter association of CD150 with SHIP, but the amount of SHIP associated with CD150 does not depend on SH2D1A levels. Thus, SH2D1A does not simply compete with SHIP for binding to CD150.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 3.

A, BCR ligation down-regulates association of SHIP with CD150 but does not affect CD150 association with SH2D1A and SHP-2. Western blot of CD150 immunoprecipitates. After activation of cells with anti-IgM sera, immunoprecipitations were performed as in Fig. 2⇑. As a control for CD150 levels after BCR ligation, we included immunoprecipitation of surface-biotinylated CD150 followed by Western blot with streptavidin peroxidase. One of five experiments. B, At the same time BCR ligation did not change the level of SHIP and SH2D1A in particulate or cytosolic fractions or detergent-insoluble glycolipid-enriched domains (DIGs). The presence of Lyn served as a control for detergent-insoluble glycolipid-enriched domains preparation (lower panel). Western blot of subcellular fractions. One of three experiments. C, Regulation of SH2D1A expression via BCR and CD40. BCR ligation down-regulated and CD40 engagement up-regulated SH2D1A expression in MP-1 cells after 48 h of stimulation. Western blot analysis with anti-SH2D1A serum on NP-40 lysates. Anti-p38 mitogen-activated protein kinase (p38) blots controlled equal loading. One of three experiments.

Because CD150 expression is up-regulated on naive B cells by either CD40 or BCR cross-linking (3), we tested whether SH2D1A expression in B cells could be induced and/or up-regulated. Up to 48 h after ligation of CD40, IgM, or CD150, SH2D1A expression was not induced in cell lines BJAB, Raji, or XLP-8002. In the SH2D1A⁺ cell line, CESS, ligation of CD40 or CD150 did not change the level of SH2D1A expression (data not shown). However, 48 h of stimulation via CD40 up-regulated SH2D1A expression in MP-1 cells (Fig. 3⇑C). Cross-linking of CD150 on these cells did not affect SH2D1A levels, whereas IgM ligation down-regulated SH2D1A expression (Fig. 3⇑C).

Tyr281 and Tyr327 in the CD150ct are essential for both SHIP and SHP-2 binding

To clarify the molecular basis of SHIP vs SHP-2 binding to CD150, we constructed GST-fusion proteins of the CD150ct with single replacements of tyrosine at Y269F, Y281F, Y307F, or Y327F (Fig. 4⇓A). Since SHIP and SHP-2 binding to CD150ct is phosphotyrosine dependent (5), all fusion proteins were expressed in both tyrosine-phosphorylated (PY) and nonphosphorylated forms. The major 145-kDa protein coprecipitated with GST-CD150ct-PY and phosphorylated in the in vitro kinase assay previously was identified as SHIP (Ref. 5; Fig. 2⇑C). Mutations in any one of the tyrosines did not affect binding of SH2D1A to CD150ct (Fig. 4⇓B). However, SHIP bound to GST-CD150ct-PY, M1-PY(Y269F), and M3-PY(Y307F) in a phosphotyrosine-dependent manner (Fig. 4⇓B and data not shown). At the same time, we did not detect SHIP in the precipitates with M2-PY(Y281F) and M4-PY(Y327F) (Fig. 4⇓B). SHP-2 was also precipitated with GST-CD150ct-PY, M1-PY(Y269F), and M3-PY(Y307F) fusion proteins, and again both M2-PY(Y281F) and M4-PY(Y327F) failed to bind SHP-2 in cell lysates from SH2D1A⁺ cells (Fig. 4⇓, C and D). On the other hand, in the absence of SH2D1A (BJAB cell lysates) both M2-PY and M4-PY were able to bind SHP-2 (Fig. 4⇓D), indicating that SH2D1A and SHP-2 are competing not only for Y281 but also for Y327. The fact that M2-PY and M4-PY bind more SH2D1A than SHP-2-binding mutants also may reflect competition for the same binding sites. These results suggest that the same tyrosines within TxYxxV/I motif in CD150ct (Y281 and Y327) are required for SHIP and SHP-2 association with CD150.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 4.

A, Replacements of tyrosine by phenylalanine in GST-fusion protein constructs of the CD150ct. B, Y281 and Y327 in CD150ct are involved in SHIP recruitment. Nonphosphorylated and tyrosine-phosphorylated GST fusion proteins of CD150ct were used for precipitations from MP-1 cells followed by in vitro kinase assays on precipitates (SHIP) and Western blot with anti-SHIP (data not shown) and anti-SH2D1A serum. The tyrosine-phosphorylated fusion proteins GST-M2-PY (Y281F) and GST-M4-PY (Y327F) did not precipitate SHIP. One of five experiments. C, D, Both Y281 and Y327 in the CD150ct are involved in SHP-2 recruitment. Western blot analysis of fusion proteins precipitates with anti-SHP-2 and anti-SH2D1A Abs. Tyrosine-phosphorylated fusion proteins were used for precipitations from MP-1 (C and D, upper and middle panels) and BJAB (D, lower panel) cell lines. The tyrosine-phosphorylated fusion proteins GST-M2-PY (Y281F) and GST-M4-PY (Y327F) did not precipitate SHP-2 from the MP-1 cell line in the presence of SH2D1A. At the same time, mutations in Y269 and Y307 (fusion proteins GST-M1-PY and GST-M3-PY did not affect binding of SHP-2 to CD150ct (upper panels of C and D). Recruitment of SH2D1A to CD150ct was revealed by Western blot with anti-SH2D1A serum. In the absence of SH2D1A (cell line BJAB), GST-M2-PY (Y281F) and GST-M4-PY (Y327F) were able to recruit SHP-2 (D). Tyrosine phosphorylation of fusion proteins were controlled with anti-PY mAb 4G10 (D, lower panel). One of four experiments.

This study provides evidence of SH2D1A expression in human B cells. Using in vivo and in vitro approaches, we showed that both SHIP and SHP-2 are able to bind CD150 in B cells. Point mutations of tyrosines in GST-fusion proteins of the CD150 cytoplasmic tail revealed that tyrosine phosphorylation of the same residues Y281 and Y327 are essential for SHIP as well as SHP-2 binding. Replacements at residues Y269 and Y307 did not affect either SHIP or SHP-2 association with CD150. How then is SHIP and SHP-2 binding to the same sites regulated? Our data indicate that the adaptor protein SH2D1A is involved in this regulation. Despite the reported absence of SH2D1A expression in murine B cells (11), SH2D1A mRNA (12, 13) and protein are expressed in human B cells, including the B cell lines MP-1 and CESS and tonsillar B cells (Fig. 1⇑). As in T cells, CD150 coprecipitates with SH2D1A in B cells, and neither phosphorylation nor mutations of tyrosines in CD150 affect this association (Figs. 2⇑ and 4⇑). The SH2 domain of SH2D1A binds the sequence surrounding Y281 devoid of tyrosine (15, 16) and also binds in a phosphotyrosine-dependent manner to the 2B4 receptor (17). Furthermore, SH2D1A can compete with SHP-2 for binding to the cytoplasmic tail of CD150 and 2B4 (10, 17). Here we found that in SH2D1A-expressing B cell lines, CD150 coprecipitates with SHIP; however, in SH2D1A⁻ cell lines, CD150 associates only with SHP-2, and the presence of SH2D1A facilitates SHIP binding to CD150ct (Fig. 2⇑C). Probably, the regulation of CD150 association with SHIP vs SHP-2 by SH2D1A may be under control of BCR and CD40 signaling since: 1) short term signal via BCR decreases SHIP association with CD150 (Fig. 3⇑A); and 2) long term (48 h) BCR ligation reduces and CD40 cross-linking up-regulates the level of SH2D1A expression (Fig. 3⇑C).

Modeling of the CD150ct

The cytoplasmic tail of CD150 has the paired tyrosine-based motif TxYxxV/I, which we propose to be designated as a “switch” motif (ITSM). This motif with the help of the adaptor protein SH2D1A may control binding of tyrosine vs inositol phosphatases to receptors. This motif is different from the well-defined immunoreceptor tyrosine-based activation motifs D/ExxYxxL/I(x)6-8YxxL/I in BCR and TCR complexes, which on phosphorylation recruit protein tyrosine kinases such as Syk and ZAP-70 (39, 40). However, the CD150/2B4 motif has some similarities with immunoreceptor tyrosine-based inhibitory motifs (ITIMs) I/VxYxxL/V(x)26-31I/VxYxxL/V found within cytoplasmic domains of “inhibitory receptor superfamily” members such as FcRIIb, CD22, CD72, killer Ig-related receptor, paired Ig-like receptors, p49B, Ig-like transcript (ILT), and leukocyte-associated Ig-like receptor (41). These ITIMs inhibit activation receptors by recruiting SH2-containing tyrosine phosphatases SHP-1 and SHP-2, and also SHIP (1, 41, 42).

We performed a series of protein sequence databases searches to broaden the list of ITSM-containing molecules. Most of the previously reported CD150 homologues, 2B4 (CD244), CD48, CD84, and Ly9 (CD220), belong to CD2 subfamily of the Ig superfamily. Multiple alignments of CD2 subfamily members with the highest level of homology to CD150 (2B4, Ly9, CD84) were used to build HMM profiles for position-specific matching searches against a Swiss-Prot database with a HMMER program package. Because we consider the ITSM motif the main functional unit within the CD150ct, the key criterion for sequence selection was the presence of tyrosine-based motifs that fit the ITSM consensus (Fig. 5⇓A). The common feature for members of the CD2 subfamily is a paired ITSM and the existence of differentially spliced truncated forms leading to only a single ITSM for CD150, 2B4, and Ly-9 (Fig. 5⇓, A and B).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 5.

A, Multiple alignments of CD150 TxYxxV/I motifs (ITSM) and the tyrosine-containing motifs from the sequences found by HMM searches. B, Schematic chart representing the distribution of ITIM and ITSM-like motifs in the cytoplasmic domains of the molecules with homology to CD150. □, ITIM; ▪, ITSM. PD-1, programmed death-1 receptor. C, Ribbon diagram illustrating the three-dimensional structure of CD150ct. The diagram was prepared using a Swiss PDB Viewer.

Search results for CD150, 2B4, CD84, and Ly9 alignment revealed homology to several members of SHPS-1 (SHP-2 substrate 1) family: SHPS-1, BIT, and MYD. These sequences are highly homologous to each other and contain two ITSM-like motifs alternated with ITIM consensus motifs (Fig. 5⇑, A and B). The cytoplasmic tail of SHPS-1, like CD150, is able to bind SHP-2 phosphatase.

The third most represented group of sequences belongs to the Siglec/CD33 family. The cytoplasmic domains of CD31/platelet endothelial cell adhesion molecule-1, CD33, Siglec-5 (OB/BP2), and Siglec-9 all have a similar tyrosine-based motif distribution pattern. A ITSM-like motif is situated 3–7 residues from the C terminus and is preceded by a conventional ITIM motif. Interestingly, similar to CD150ct, a GST platelet endothelial cell adhesion molecule/CD31 cytoplasmic tail can bind both SHP-2 and SHIP, and SHIP interacts predominantly with the ITSM motif in CD31 (43).

CEA superfamily members were also widely represented in the retrieved sequences with ITSM motifs. Bgp-1, Bgp-2, C-CAM 105 ecto-ATPase, and related molecules contain Y-based motifs the sequences and positions of which in the cytoplasmic domain are highly similar to the positions and sequences of ITSMs: tyrosines at the C terminus followed by a group of positively charged residues. Similar to CD150, which recently was shown to serve as alternative measles virus receptor (44), both Bgp-1 and Bgp-2 are receptors for mouse hepatitis virus and also have truncated forms (45).

Unlike sialoadhesin and CEA family members, other ITSM and ITIM-containing molecules retrieved by the HMM search demonstrate experimentally confirmed inhibitory activity. CD150 showed a weak sequence homology to several members of the recently established monocyte-inhibitory receptor/LIR/ILT family, like paired Ig-like receptors B and programmed death-1 (PD-1) receptor. However, ITIMs rather than ITSMs are presented more widely in the cytoplasmic domains of this group, and motif homology to CD150 ITSMs is weaker than in the case of Siglecs or CEAs (Fig. 5⇑, A and B). Nevertheless, the structural similarity between these sequences and CD150 creates a potentially important bridge between these two groups of molecules. ITSM-containing molecules also include the catalytically inactive tyrosine kinase human receptor related to tyrosine kinase and the membrane adaptor protein called SHP-2-interacting transmembrane adaptor protein (SIT) (46) (Fig. 5⇑, A and B). The effector molecules that bind the ITIM motif in SIT have not been identified, but it is likely that SIT regulates TCR-mediated induction of IL-2 gene transcription via this motif (47).

The distribution of ITSM-like tyrosine-containing motifs in CD150 homologues confirms that these motifs represent an important and structurally unique but poorly studied group of tyrosine-containing regulatory motifs. Multiple alignment of ITSM-like motifs from distant homologues of CD150 indicate the conservation pattern of this kind of motif: 1) in most of the motifs presented on Fig. 5⇑A, at least one additional conserved position is evident: a threonine or serine in position −4 with respect to the ITSM tyrosine. This pattern is present in all ITSMs (Y281 and Y327 in CD150, 2B4, CD31) that have been shown to be functionally significant; 2) in members of CD2, Siglec, and CEA families, one ITSM is positioned 3–5 residues upstream of the C terminus; 3) CD150, Ly9, 2B4, and murine Bgp-2 have alternatively spliced forms, and shorter isoforms possess only a single ITSM; 4) in several families (SHPS, Siglec, and LIR, both ITIM and ITSM consensus motifs are present.

We applied computational biology methods to explore possible mechanisms of CD150ct interactions with associated molecules and to explain the available experimental data. No sequences with significant homology to CD150 were found in Protein Data Bank (PDB) by either search algorithm we used, including the position-specific scoring methods (30, 48); therefore, homology modeling was inapplicable. Threading of the CD150ct sequence by the H3P2 method predicted an Ig-like β sandwich fold for CD150ct. One structure (IgG1 H chain, PDB access code 1tet, residues 113–213) with the best z score and alignment to CD150 was selected for modeling. The model we built represents a Greek key fold, typical for Ig-like domains, with two parallel β sheets that contain two and four strands (Fig. 5⇑C). Checking the quality of the resulting structure by ERRAT (36) and Verify3D (37) programs indicated that the modeled structure might be natural. Realizing the limitations of the used modeling method, nevertheless, we can use this model for analysis of the experimental data from a structural perspective and make functional predictions.

Since Y281 and Y327 are essential for SHP-2 binding to CD150, it is possible that both SH2 domains of SHP-2 mediate this association. Using our model and SHP-2 structure revealed by crystallography (49), we predicted that simultaneous binding of Y281 and Y327 by two SHP-2 SH2 domains is unlikely, because the distance between tyrosine-binding pockets of N- and C-terminal domains of SHP-2 is 40 Å, compared with 8.18 Å, between Y281 and Y327 residues in CD150ct model. Moreover, it was shown that only the N-SH2, but not the C-SH2 domain of SHP-2 phosphopeptides binds to both ITIM- and ITSM-like motifs of CD31 (43).

Apparently, displacement of SHP-2 by SHD1A (10) makes Y281 and Y327 available for binding by SHIP, which also requires Y281 and Y327 (Fig. 4⇑B). This possibility is supported by in vivo data demonstrating differential coprecipitation of CD150 with SHIP in SH2D1A⁺ cell lines or with SHP-2 in SH2D1A⁻ cell lines (Fig. 2⇑). Although SHIP possesses only one SH2 domain, both Y281 and Y327 in CD150 are important for association with SHIP (Fig. 4⇑). Similarly, the association of SHIP with FcγRIIB also requires two tyrosines; not only Y279 within the ITIM of the FcγRIIB is required, but also Y296 is necessary for stable association of SHIP with FcγRIIB (50). Interestingly, that SH2 domain of SHIP has a much higher affinity to immobilized phosphopeptide of CD31 ITSM-like motif, than N-SH2 domain of SHP-2 (43). However, using only surface plasmon resonance with evaluation of dissociation constants for SH2 domains and their affinity to ITSMs in CD150, we will be able to build a kinetic model of their competitive binding.

Why should SH2D1A binding to the sequence surrounding Y281 in CD150 prevent its association with SHP-2 but not compete with SHIP? Amino-terminal residues adjacent to Y281 in the cytoplasmic tail of CD150 are critical for high affinity binding of SH2D1A to CD150 (16). At the same time, mutational analysis of the FcγRIIB tail revealed that the residue Y279 within the ITIM motif is required for SHP-2 but not SHIP binding to the cytoplasmic tail of FcγRIIB (50, 51). Therefore, high affinity binding of SH2D1A to residues amino-terminal to Y281 in CD150 may compete with SHP-2 but not affect association of SHIP with CD150. Taken together, SH2D1A may play a role as a molecular “switch” that regulates SHIP vs SHP-2 association with CD150. This model is consistent with the reported involvement of SHIP in regulation of B cell development as well as immune responses to antigenic challenge (52, 53, 54). In other words, CD150 may transmit SHIP-dependent or SHP-2-mediated signals at distinct stages of B cell maturation, and the adaptor protein SH2D1A may regulate this switch. Using DT40 sublines, we are currently investigating what CD150-mediated signal transduction pathways are regulated by SH2D1A. After this paper was submitted, Nagy et al. (55) reported the expression of SH2D1A in B cell lines derived from EBV-positive Burkitt’s lymphomas.